Drugs and the pharmaceutical sciences

470 Gottschalk
FIGURE 4 Comparison of (A) dead-end (normal flow) and (B) tangential (cross-flow) filtration.
In each panel, the large arrow shows the direction of feed flow and the small arrows show the
direction of permeate accumulation.
only used where the retentate load in the feed stream is anticipated to be low (e.g., sterile
filtration for product filling, virus removal). Where filtration is used for clarification and
size fractionation, tangential-flow filtration (also known as cross-flow filtration) is
preferred. In this configuration, the feed flow is parallel to the filter medium and thus
perpendicular to the flow of permeate. This allows retained species to be swept along the
filter surface and out of the device, helping to maintain high flux levels even with large
amounts of retentate.
Tangential-flow filter modules come in many designs. These differ in terms of
channel spacing, packing density, cost, pumping energy requirements, plugging tendency
and ease of cleaning, so the design must be chosen on a case-by-case basis for each
bioprocess depending on the implications of the above criteria in the context of the
overall process train (Jornitz et al., 2002; Dosmar et al., 2005). Filter media can be
divided into two major types: surface filters and depth filters (Fig. 5A). Surface filters are
essentially thin membranes containing capillary-like pores. Particles or molecules which
are too big to pass through the pores are retained on the membrane surface, that is, the
retentivity is “distinct” at a certain particle-size in regard to the pore size/cut-off and this
can be validated as discussed below.
In contrast, depth filters have a thicker ‘bed’ of filter medium rather than a thin
membrane, and particles are trapped in the interstices of the internal structure, which
describes a torturous path from one side of the filter to the other. To increase the surface
area available for filtration without increasing the footprint, depth filter pads are either
inserted manually into filter presses or, more preferably, supplied as lenticular filters, in
which multiple filter pads are pre-assembled in a modular housing.
The materials used to construct depth filters include cellulose fibers, inorganic filter
aids, resin binders and synthetic polymers, offering a large inner surface area and void
volumes of up to 85% (Singhvi et al., 1996; Prashad and Tarrach, 2006). Inorganic filter
aids such as diatomaceous earth and perlite increase the permeability and retention
characteristics of the filter matrix, while synthetic polymers and resin binders increase
the strength of the filter medium and generate a net positive charge that helps to trap
colloids. For these reasons, depth filters can trap particles much smaller than the
maximum pore size. The retention mechanism of depth filters is not absolute, and
changes during operation as retentate builds up in the filter matrix. Therefore, depth
filters cannot be validated for use in sterile filtration in the same way as membrane filters,
but because they are less expensive than membrane filters they are often employed in prefilter
steps to remove cells and debris. For example, feed streams from mammalian cell